RESUMO
All fields of science depend on mathematical models. Occam's razor refers to the principle that good models should exclude parameters beyond those minimally required to describe the systems they represent. This is because redundancy can lead to incorrect estimates of model parameters from data, and thus inaccurate or ambiguous conclusions. Here, we show how deep learning can be powerfully leveraged to apply Occam's razor to model parameters. Our method, FixFit, uses a feedforward deep neural network with a bottleneck layer to characterize and predict the behavior of a given model from its input parameters. FixFit has three major benefits. First, it provides a metric to quantify the original model's degree of complexity. Second, it allows for the unique fitting of data. Third, it provides an unbiased way to discriminate between experimental hypotheses that add value versus those that do not. In three use cases, we demonstrate the broad applicability of this method across scientific domains. To validate the method using a known system, we apply FixFit to recover known composite parameters for the Kepler orbit model and a dynamic model of blood glucose regulation. In the latter, we demonstrate the ability to fit the latent parameters to real data. To illustrate how the method can be applied to less well-established fields, we use it to identify parameters for a multi-scale brain model and reduce the search space for viable candidate mechanisms.
Assuntos
Biologia Computacional , Aprendizado Profundo , Biologia Computacional/métodos , Humanos , Redes Neurais de Computação , Glicemia/metabolismo , Algoritmos , Modelos Teóricos , Modelos BiológicosRESUMO
The brain primarily relies on glycolysis for mitochondrial respiration but switches to alternative fuels such as ketone bodies (KBs) when less glucose is available. Neuronal KB uptake, which does not rely on glucose transporter 4 (GLUT4) or insulin, has shown promising clinical applicability in alleviating the neurological and cognitive effects of disorders with hypometabolic components. However, the specific mechanisms by which such interventions affect neuronal functions are poorly understood. In this study, we pharmacologically blocked GLUT4 to investigate the effects of exogenous KB D-êµ-hydroxybutyrate (D-êµHb) on mouse brain metabolism during acute insulin resistance (AIR). We found that both AIR and D-êµHb had distinct impacts across neuronal compartments: AIR decreased synaptic activity and long-term potentiation (LTP) and impaired axonal conduction, synchronization, and action potential properties, while D-êµHb rescued neuronal functions associated with axonal conduction, synchronization, and LTP.